MQA vs HiRez: an apples-to-apples comparison - FINAL

[Miska]

On 10/30/2018 at 9:04 AM, manisandher said:

I'm still totally done with all hires. It seems to me that if you can't get properly done hires from ECM/Qobuz, then you're not likely to get it from anywhere. And no wonder MQA sounds strange to some people, if most of it has been derived from 'fake' hires.

 

I have a lot of good hires from highresaudio.com, nativedsd.com and 2L (and some from HDtracks too).

 

RedBook is always compromised and stuffed, and needs a lot of careful DSP massaging to get it fixed and something proper out of it.

 

By definition, you must not artificially band limit the source. And with RedBook, pretty much anything else than solo piano you end up band limiting. And you irrevocably destroy part of the sound.

 

MQA in turn is screwed up in so many technical ways and has a lot of other unwanted features. But if someone likes to listen to a distortion generator called MQA, I don't have problem with it. But I just avoid it.

 

Just my opinion...

[Miska]

3 hours ago, PeterSt said:

Looks odd to me after all. Something now is wrong. Including the rising slope beyond that point.

 

The noise slope is typical and you can see start of it on most 192/24 recordings. It is just the SDM ADC's noise slope.

 

Usually you can find low level spurious tone around 30 - 35 kHz which is SMPS switching frequency.

 

The 44.1k line is probably word clock used to synchronize converters leaking at low level to the ADC.

[Miska]

4 hours ago, manisandher said:

Which instrument has energy right up to 96 kHz (!)?

 

https://www.cco.caltech.edu/~boyk/spectra/11.htm

https://www.cco.caltech.edu/~boyk/spectra/13.htm

 

Important thing to note is that sampling rate should be high enough so that any instrument content has spectrum fallen below the background noise level by Nyquist frequency of the sampling rate. This avoids adverse effects of band limiting.

 

For making wide bandwidth recordings, you also need wide bandwidth microphones. The best I'm aware is this:

http://www.sanken-mic.com/en/product/product.cfm/3.1000400

Some more info, including recordings made with it:

http://www.sanken-mic.com/en/report/reports.cfm?top=1&id=12

 

2L uses DPA microphones (ex B&K), like:

https://www.dpamicrophones.com/ddicate/4007-omnidirectional-microphone

which reach up to 50 kHz. There used to be hard-core variant of these with 130V phantom power, many use those ones.

 

Sennheiser has 8000-series which has ones going a bit higher:

https://en-fi.sennheiser.com/recording-condenser-microphone-onmi-directional-guitar-acoustic-bass-brass-mkh-8020

And also:

https://en-fi.sennheiser.com/studio-condenser-microphone-digital-recording-systems-mkh-800-p48

 

[Miska]

1 hour ago, manisandher said:

Looks like the 88.2 sample-rate Nyquist line to me. Take a look at a close up of one of the peaks:

 

 

Surely that's clear imaging right there?

 

I'd say it is nulls of the PCM decimation filter...

 

[Miska]

1 hour ago, manisandher said:

 

And this? (2L's Haydn String Quartet in D Op 76 No 5 - Finale - Presto - DXD):

 

 

What a total mess! And this from 2L!

 

Mani.

 

If you want something from 2L that has flat noise floor (modulator noise completely filtered out), take a newer recording that has been made using Merging Horus.

 

[Miska]

5 minutes ago, manisandher said:

Did this use the Horus? If so, I'm not impressed - there's as much noise at 176.4 kHz as there is actual signal between 3-6 kHz:

 

 

Why would this recording require anything more than redbook?

 

Well, that would need at least 88.2k sampling rate.

 

But by quick check following are from Horus:

2L-109

2L-110

2L-111

2L-125

2L-139

 

And then of course these:

http://www.2l.no/hires/DXD-DSD/index.html

 

[Miska]

20 minutes ago, manisandher said:

But there's virtually no signal above 22.05 kHz.

 

To me it looks like the content fades to background noise around 40 kHz. But you have very little scale on X-axis so it is hard to say.

 

For fun I ran some tests for Mark Knopfler's Privateering and David Gilmour's Rattle That Lock. Also Knopfler's Tracker (in 192/24) showed similar content.

 

Of these, the first "Don't forget your hat" is clearly band limited by ADC's decimation anti-alias filter. And some parts of "Faces of stone".

[Miska]

54 minutes ago, manisandher said:

More like 26 kHz. But the signal peaks at around -125dB at 22.05 kHz:

 

But just look at all that HF noise you'd be putting through your system! For what?

 

'JGH_take-111_raw-DXD' seems perhaps worthy of a higher-than-redbook sample rate:

 

 

Though the signal above 22.05 kHz remains vanishingly small. At least the HF noise is under control with the Horus.

 

From the older recordings I have found content up to 60 kHz.

 

Horus just runs more aggressive filter for the modulator noise, you can see it begins to roll-off around 88.2 kHz. While the older one (IIRC, AX24) runs less filter and thus has also some aliased noise at the top of the spectrum. Not so different compared to for example AKM chips running at 352.8k:

 

Remember that all modern ADCs output DSD-like data that is just converted to PCM inside the ADC chip using digital decimation filter.

 

For comparison, same ADC, but instead of PCM, running at DSD256 and then software converted to 352.8k PCM with standard compliant noise filter:

And alternative noise filter:

 

[Miska]

30 minutes ago, manisandher said:

Getting back to the track at hand, here's the spectrum/spectrogram of the first minute of Sample B (original hires), before the cymbals kick in:

 

Is it possible to record an instrument like this...

 

... and produce output that extends up to 48 kHz?

 

If the answer is 'yes', then all may be good with the track.

 

Yes, likely it is, with that kind of mic setup. Not the instrument itself after you have released the string, but the sound of nails hitting the strings can produce fairly sharp transients which you would see as horizontal lines in your vertical flowing spectrogram.

 

You could of course try it yourself. I purchased bunch of percussive instruments in the back to try these things out myself. Most expensive investments are wideband microphones:

https://www.thomann.de/gb/sennheiser_mkh_8020_stereo_paar.htm

 

But what does this have to do with MQA?

[Miska]

1 hour ago, manisandher said:

It's in the title of the thread - 'MQA vs. hires'. I'm trying to establish whether the track I used in this thread is a 'legit' example of hires, after earlier comments such as:

 

OK, not sure about that one, since I haven't looked. I've looked into material I have. I've seen some material, also in MQA that are clearly upsamples. Some material is combination of upsampled and real hires, one such example being Pink Floyd's Endless River album. The older material on that album is recorded in 48k and newer in 96k. So for mixing they've upsampled the older 48k material to 96k.

 

I can imagine also new material where you could find mixture of sampling rates, because some gear have digital output at 48k (including some digital mics, like some wireless ones) and you could be still recording other stuff in analog at 96k. And with good ADC, even if you record it from analog outputs of a synth or something like that, at 96k, you may capture features of the not so perfect DAC in the synth or other piece of digital or semi-digital gear...

 

So I'd say this is a complex question sometimes.

 

[Miska]

I just don't understand what people see/hear in ECM's productions. Nothing I would want to, for any reason...

 

[Miska]

4 hours ago, manisandher said:

I've spent the last couple of days listening to hires vs. redbook I have from the same master, and can only say that I'm generally totally fine with the sound of the redbook. But I'm using PeterSt's software player and DAC, which have been specifically designed to optimise redbook (though not at the expense of hires).

 

To me it sounds strange if original format sounds worse than the down-conversion... That would mean applying a brickwall filter at 22.05 kHz and dithering down to 16-bit makes it sound better or at least as good on your system. That also tells something.

 

[Miska]

27 minutes ago, sapporo said:

Best microphones go up to 50 kHz. Room/loudspeaker systems filter around

20 kHz. Mature male ears cut off at 15-16 kHz. Physics limits DACs to ca 20 bit resolution. No wonder most of us find Redbook good enough as a distribution codec.

 

Best microphones go up to 100 kHz. My loudspeakers and headphones go to 50 kHz. And remember that the response doesn't end there in a brickwall, but instead just have a downward slope.

 

Anyway, I rarely listen through loudspeakers anyway. I almost always listen through headphones.

 

42 minutes ago, manisandher said:

The only cases where the originals obviously sound better than the redbooks to me are some of the 2L DXDs. I suspect you share the same experience.

 

I have plenty of hires material that sounds better than RedBook... Some sound worse because they are just poor upsamples. I think I have only one hires album that is not upsample but sounds worse in hires than in RedBook, but could have different mastering then too. But even that could also be because RedBook is just hiding flaws of the recording. A bit like MP3 can "fix" some bad recordings by removing some of the distortions. Probably MP3/AAC would also "fix" some of the MQA's aliasing flaws too.

 

[Miska]

24 minutes ago, sapporo said:

I am really curious about the 100kHz mics and 50 kHz speakers with the typical 0.5 dB tolerance for the former and 3dB for the latter. Could you give the model names please? In case of speakers I wrote speaker/room system, but ok, let it be anechoic specs. Rooms attenuate highs. But you are not equipped with above 20 kHz hearing

I presume?

 

I listed mics earlier in this thread. But one of the mics was:

http://www.sanken-mic.com/en/product/freqpola.cfm/3.1000400

As you can see, the response is also diffuse-field compensated, so it gives pretty even frequency response in a normal space.

 

Also Bruel&Kjaer has wideband microphones, like the 4939 model that is 4 Hz to 100 kHz +-2 dB. 4138 model goes 6.5 Hz to 140 kHz, but is a bit low on sensitivity.

 

I use Elac loudspeakers with their JET tweeters (AMT ribbon). Planning to get Monitor Audio Studio's to the office too. But that is anyway about 5% of my listening. Rest is with various headphones.

 

[Miska]

3 hours ago, sapporo said:

Bruel&Kjaer are measurement mics.

 

So? Microphone doesn't know what it is capturing. As listed earlier, DPA and Sennheiser also have recording mics that go to 50 - 60 kHz.

 

3 hours ago, sapporo said:

The  Monitor Audio Studios have mixed reviews, but again, it's  a matter of taste and preferences.

 

I don't care about review, What Hifi the least. I have heard those speakers myself, with my own ears. That's why I'm considering them...

 

3 hours ago, sapporo said:

AMT ribbon tweeters-- forgot about those. As many pros as cons. In the end it's also a taste thing, I dislike the sound.

 

I cannot stand metal domes, most likely due to the strong resonance just above 20 kHz. Painful to listen. For quite some time I've been preferring speakers with folded or non-folded ribbon tweeters.

 

3 hours ago, sapporo said:

unless the listening position is exclusively near-field

 

I think my headphones are fairly near field. For loudspeaker listening I have 2 meter distance to speakers, with speakers of course toed in, and calibrated through digital room correction for the listening position.

 

This is not about hearing the high frequency components in isolation, but instead hearing their influence to the transient wavelet shape.

 

[Miska]

2 hours ago, sapporo said:

My bad. Of course a human cannot perceive the highs above 20 kHz firstly due to the auditory apparatus construction, never mind the attenuation by air or walls.

I must have been still thinking about the audiophile  bats.

 

http://www.ica2016.org.ar/ica2016proceedings/ica2016/ICA2016-0219.pdf

https://www.researchgate.net/publication/245524994_Hearing_threshold_for_pure_tones_above_20_kHz

https://www.physiology.org/doi/full/10.1152/jn.2000.83.6.3548

 

With RedBook you have a brickwall filter below frequencies many people can hear even with pure tones. Having such strong transition in such place is just plain bad.

 

[Miska]

2 minutes ago, sapporo said:

• Agree. I use silk dome tweeters.

 

In living room I have Dynaudio speakers with silk domes. Very nice sound, rolls off smoothly above 25 kHz and resonance frequency is below cross-over frequency. Their silk domes with high power handling capability also allow lower than usual cross-over frequency.

 

[Miska]

36 minutes ago, sapporo said:

The above quoted acoustics research says that a small percentage of people  can "hear" ("feel" is a better word, mechanism unknown)  pure tones at SPL>90 dB up to 24kHz. This is old hat and  has nothing to do with reproduction of music.  The 48kHz codec  covers even that most bizzarre of audiophile scenarios. Pump the 24khz pure tones'  volume up to 130dB and maybe half of population will "hear", or die trying.

 

I wouldn't say 4 out if 15 listeners is small percentage. And 15 is really small sample. And the test was limited to 100 dB SPL which is not particularly high, something typical symphony orchestra will put out. Think how many out of million that would be? How many of that "small" percentage out of million would be audiophiles? Chances are high that quite big. And hearing is very heavily learning-based. These were still about hearing pure tones, tones combined with base band fundamental are again a different thing. I have experienced that myself after spending years teaching passive sonar operators to listen underwater sounds to detect submarines and such (and developing systems for that). Brain is extremely capable and powerful detector that can compensate for lot of defects in audio delivery paths, including hearing. It takes some time and effort to become good listener, but I'd say most audiophiles are way above average on that front!

 

Not really. In fact 48 kHz sampling rate has exactly the nasty ringing at 24 kHz. Another aspect is that the ringing or leaky filters coincides exactly with resonance frequency of aluminum dome tweeters where you have typical 20 dB boost....

 

You really need to go for at least 88.2 kHz sampling rate or more. IMO, the only tolerable case is that Nyquist frequency is above the point where microphone input naturally falls below background noise. And with good microphones that is well past 50 kHz. (microphone output reaching about -120 dB level) When you start band-limiting, you wreak havoc already. For long time I've had opinion that DSD is best way at the moment to deliver this.

 

It doesn't even need much extension to already become a lot better. If you listen at many of the remasters/productions made by Steven Wilson, all those look like relatively (for digital filter) slow roll-off band limited at 30 kHz, most likely by Lavry ADC, the difference is already huge. Most recent, only couple of days old, one:

https://www.highresaudio.com/en/album/view/tf263a/steven-wilson-home-invasion-in-concert-at-the-royal-albert-hall-live

I can only wonder how good those could become without any artificial band limiting!

 

[Miska]

5 minutes ago, manisandher said:

Vibrant. Colour. More bite. Too 'etched' - like an artificial sheen/halo around instruments. More interesting to listen to, but fatiguing in the long run.

 

Usually when I hear such I think there's coloration or distortion.

 

So how about taking the hires and putting your own 30 kHz filter on it? Why would you accept someone's choices except yours and take the original instead?

[Miska]

14 minutes ago, sapporo said:

I really have no problem with hi-res audio for home users. And  I still have  a collection of SACDs, gathering dust in times of Tidal.

I just think that there are so many more important factors in sound quality of music reproduced at home. First is loudspeakers and the room. Moving to good speakers and doing some acoustic adaptations can be a paradigm shift in home music reproduction. Then the quality of the master. I find many tracks from 50-60 years ago much better than a vast majority of today's releases, hi-res or no. Then the electronics, but with  a smaller impact than the previous two.  And of course, first of all, good music and the company to share the experience with. 

 

I'm not arguing with any of that. I use digital room correction, etc. One reason I started developing my own player was that I wanted proper digital room correction also for DSD content. Hires is just part of the solution. But there's even quite a bit of new material available even in DSD256.

 

Speaking of old recordings, my favorite variant of Jazz at the Pawnshop is the DSD128 version. Although DXD one is very good too.

 

And one of the most moving old recordings is Pink Floyd's Live at Pompeii, the original one.

 

[Miska]

27 minutes ago, sapporo said:

Jazz at the Pawnshop was recorded with  Nagra IV at 38 cm/s (second machine for reel changes).  Two main mics  Neumans U47 using ORTF setup, some extra mics.  This recording ( I have the SACD)  inspired me to get my own  Nagra Seven and two AKG C414. 

Jazz at the Pawnshop - what's the big deal? | Steve Hoffman Music Forums

 

Given the age, the recording is pretty amazing. You can check tech specs tab about the new transfers from the tape:

https://naxos.nativedsd.com/albums/jazz-at-the-pawnshop

 

I have both DXD and DSD128 versions.